In case of recording a digital information signal, a digital modulation called a channel coding is performed to raise a recording density on a recording medium and to reduce the DC component of a transmission signal.
A Miller modulation is known as a conventional channel coding. However, the Miller modulation cannot reduce the DC component of the modulated output to zero. For example, in case of recording an audio PCM signal by a rotary head, a recording signal is transmitted through a rotary transformer. If the DC component is included in this recording signal, since the rotary transformer cannot transmit the DC component, there will be caused a problem such that the waveform of the recording signal is distorted. In the case where many low frequency components are included in the recording signal, it is necessary to extend the frequency characteristic of the rotary transformer to a low frequency band. Further, in case of using a recording method whereby the crosstalk between the adjacent tracks is suppressed by making the extending direction of the gap in the recording head differ with regard to the adjacent tracks, the effect of suppressing the crosstalk with respect to the low frequency component of the recording signal becomes insufficient.
In case of reproducing an optical disc whose surface is coated by a reflecting layer, DC noises are caused due to the dust adhered on the reflecting layer, scratches on the reflecting layer or the like. If the digital signal recorded on the disc includes no DC component, such DC noises can be eliminated by a filter.
From the above viewpoint, it is required to reduce the DC component of the recording signal to zero. For this purpose, a modified Miller (referred to as M.sup.2) modulation has been proposed. Further, a modulating method (referred to as M3 modulation) having the effect of suppressing the DC component even more than the M.sup.2 modulation has also been proposed.
In these M.sup.2 and M3 modulating methods, the minimum transition interval of the modulated output is T (where, T is a length of one bit cell) and the maximum transition interval is 3T. It is desirable that the maximum transition interval, namely, the maximum value of the interval between data value transition, be short in order to reduce the low frequency component and to improve the clock recovery capability in a reproducing circuit.